A Drosophila model of mucopolysaccharidosis IIIB

Abstract

Mucopolysaccharidosis type IIIB is a rare lysosomal storage disorder caused by defects in alpha-N-acetylglucosaminidase (NAGLU) and characterized by severe effects in the central nervous system. Mutations in NAGLU cause accumulation of partially degraded heparan sulfate in lysosomes. The consequences of these mutations on whole-genome gene expression and their causal relationships to neural degeneration remain unknown. Here, we used the functional Drosophila melanogaster ortholog of NAGLU, Naglu, to develop a fly model for MPS IIIB induced by gene deletion (Naglu^KO), missense (Naglu^Y160C), and nonsense (Naglu^W422X) mutations. We used the Drosophila activity monitoring system to analyze activity and sleep and found sex- and age-dependent hyperactivity and sleep defects in mutant flies. Fluorescence microscopy on mutant fly brains using Lysotracker dye revealed a significant increase in acidic compartments. Differentially expressed genes determined from RNA sequencing of fly brains are involved in biological processes that affect nervous system development. A genetic interaction network constructed using known interacting partners of these genes consists of 2 major subnetworks, one of which is enriched in genes associated with synaptic function and the other with neurodevelopmental processes. Our data indicate that lysosomal dysfunction arising from disruption of heparan sulfate breakdown has widespread effects on the steady state of intracellular vesicle transport, including vesicles associated with synaptic transmission. Evolutionary conservation of fundamental biological processes predicts that the Drosophila model of mucopolysaccharidosis type IIIB can serve as an in vivo system for the future development of therapies for mucopolysaccharidosis type IIIB and related disorders.

Sanfilippo syndrome, rare genetic disease, lysosomal storage disorder, NAGLU, sleep fragmentation, transcriptomics, RNA sequencing

Introduction

Mucopolysaccharidosis IIIB (MPS IIIB), also known as Sanfilippo syndrome type B, is a rare lysosomal storage disorder with an incidence of around 1 in 100,000 births (Zelei et al. 2018). It is caused by mutations in the alpha-N-acetylglucosaminidase (NAGLU) gene (Zhao et al. 1996). NAGLU encodes a lysosomal enzyme required for the stepwise degradation of the glycosaminoglycan heparan sulfate (HS) (Benetó et al. 2020). NAGLU has 3 domains, with a catalytic cleft located between domains II and III. The enzyme is predominantly expressed as a “propeller-shaped” trimer with catalytic sites facing outwards. These catalytic sites cleave terminal α-N-acetylglucosamine (GlcNAc) in the nonreducing end of HS (Birrane et al. 2019). Loss-of-function mutations in NAGLU cause accumulation of partially degraded HS residues in lysosomes, which leads to cellular dysfunctions, mostly affecting the central nervous system (CNS) (Birrane et al. 2019; Benetó et al. 2020). To date, 224 NAGLU mutations have been reported, the majority of which are point mutations (Stenson et al. 2003).

HS is a ubiquitously expressed polysaccharide that serves a myriad of functions, including cell adhesion, cell signaling, cell growth, neurogenesis, immune signaling, and other roles in cellular homeostasis (Arungundram et al. 2009; Benetó et al. 2020). Extracellular HS is recycled in the lysosomes where it is degraded by several enzymes in a stepwise manner. Disruption in any of these enzymes leads to the accumulation of partially degraded HS, leading to mucopolysaccharidoses, including all MPS III subtypes (types A, B, C, and D) (Zelei et al. 2018; Benetó et al. 2020).

Individuals with MPS IIIB are born without noticeable clinical symptoms. The first signs of developmental delay appear at around 3–5 years and include impairment of language and speech skills. These are accompanied by hyperactivity and sleep disruptions, which progress to cognitive decline and gradual loss of motor functions. During adolescence, patients undergo severe neurodegeneration, resulting in loss of motor functions (Valstar et al. 2008; Benetó et al. 2020; Pearse and Iacovino 2020; Cyske et al. 2022). The lifespan of affected individuals is generally <20 years (Lavery et al. 2017).

MPS IIIB has been studied using naturally occurring or induced mutations in several model systems, including canine (Ellinwood et al. 2003), porcine (Yang et al. 2018), bovine (Karageorgos et al. 2007), avian (Palmieri et al. 2015; Genger et al. 2018), murine (Villani et al. 2007, 2009; De Pasquale et al. 2020; Petrova et al. 2023), and cellular models (Villani et al. 2009; Vallejo-Diez et al. 2018; Gaffke et al. 2020). These models have proven fundamental in understanding the roles of oxidative stress (Villani et al. 2007, 2009), neuroinflammation (Villani et al. 2007; DiRosario et al. 2009; Egeland et al. 2020; Petrova et al. 2023), changes in global metabolites (Fu, Meadows, Pineda et al. 2017) and transcriptome changes relevant to the cell cycle (Gaffke et al. 2020), and secondary storage of other substrates (Lamanna et al. 2011; Mohammed et al. 2012) as downstream effects of NAGLU mutations. Despite these advances, there is no approved therapy for this disorder (Valstar et al. 2008; Benetó et al. 2020; Pearse and Iacovino 2020; Cyske et al. 2022).

The fruit fly, Drosophila melanogaster, is a powerful model to study human disorders. Flies have a short generation time and can be reared economically in large numbers with controlled genetic backgrounds in standard laboratory conditions. Flies have conserved orthologs for ∼75% of human disease–associated genes (Pandey and Nichols 2011) and have been used to study several rare genetic disorders including MPS IIIA (Webber et al. 2018) and MPS IIIC (Hewson et al. 2024). Neuronal impairments of MPS IIIA were observed using RNAi-mediated knockdown of the N-sulfoglucosamine sulfohydrolase (Sgsh) gene, with further impairment when the autophagy products Atg1 and Atg18 were knocked down (Webber et al. 2018). Neuronal and glial-specific knockdown of the heparan-α-glucosaminide N-acetyltransferase (Hgsnat) gene revealed a potential role of glial cells in behavioral impairments in MPS IIIC (Hewson et al. 2024). Both SGSH and HGSNAT operate upstream of NAGLU in the HS degradation pathway and exhibit similar clinical manifestations when disrupted.

Naglu is a fly ortholog of NAGLU and has a Drosophila RNAi Stock Center Integrative Ortholog Prediction Tool score of 17, making it a feasible target to model the pathogenesis of MPS IIIB. We used clustered regularly interspaced short palindromic repeats (CRISPR)-induced mutations in conserved regions of Naglu orthologous to human variants. These mutations include Naglu^KO (full gene deletion) and 2 patient-derived variants, Naglu^Y160C (hNAGLU^Y140C) and Naglu^W422X (hNAGLU^W404X). hNAGLU^Y140 establishes direct contact with GlcNAc in the catalytic cleft, and the hNAGLU^Y140C mutation disrupts a critical hydrogen bond with the substrate (Birrane et al. 2019). Likewise, hNAGLU^W404X is expected to be degraded through nonsense-mediated decay or to produce a truncated polypeptide, both of which severely limit the ability of NAGLU to degrade HS. Here, using biochemical, neuroanatomical, behavioral, and transcriptional analyses, we characterize sex- and age-specific pathogenic phenotypes relevant to MPS IIIB in these Naglu mutant fly lines.

Materials and methods

Mutant generation

CRISPR-engineered mutations in Naglu were generated by WellGenetics (Taipei, Taiwan) using homology-directed repair. We received mutant (Naglu^KO, Naglu^Y160C, and Naglu^W422X) and control (Naglu⁺) flies from WellGenetics in which the second chromosomes were isogenic but the rest of the genetic background was mixed. We performed crosses to substitute the wild-type and Naglu mutant 2nd chromosomes from the WellGenetics background into the Canton-S B genetic background (Supplementary Fig. 1). The general genotype of the experimental stocks was as follows: CSBX, w¹¹¹⁸; WG2; CSB3, where WG indicates the WellGenetics control or mutant.

Drosophila culture and maintenance

Flies were reared either on NUTRI-fly molasses formulation (Genesee, Cat#: 66-123) media (used for neuroanatomical and transcriptomic assays) or on a homemade molasses formulation using the same ingredients (used for biochemical and behavioral assays) and supplemented with dry yeast. Each cross contained 5 flies of each sex and allowed 48 h for mating. Flies were maintained at 25°C, 50% relative humidity, controlled adult density, and 12-h light–dark cycle. All experiments were performed at 3–5-day-old (week 0) flies or 20–22-day-old (week 3) flies.

α-N-acetyl-D-glucosaminidase enzyme activity

α-N-acetyl-D-glucosaminidase enzyme activity was analyzed for 3 biological replicates for each genotype, sex, and age, with 20 Drosophila heads in each biological replicate. There were 3 technical replicates for each biological replicate. Samples were lysed with a bead lyser in a 50-mM citrate buffer, pH 4.5, containing 0.5% Triton X-100, and protein concentration was determined using a micro-bicinchoninic acid (BCA) assay (ThermoFisher, 23235). Prior to analysis, samples were diluted to a concentration of 0.5 mg/mL in phosphate-buffered saline (PBS) containing 0.1% Triton X-100 and 1× protein inhibitor cocktail (Sigma, cat# A32965). Six micrograms of protein lysate were added to a 20-µL reaction performed in acetate buffer, pH 4.5, containing 5 mM 4-methylumbelliferyl-N-acetyl-α-D-glucosaminide substrate (EMD Millipore, 474500). Reactions were incubated for 2 h at 37°C and quenched with 0.1 M glycine and pH 10.4. Fluorescence units were read on a BioTek Cytation 5 plate reader. Activity units were calculated relative to a 4-methylumbelliferone (Sigma, M1508) standard curve. Pairwise analyses between mutant and control were performed using a fixed effects ANOVA model on the average of all technical replicates for each biological replicate, Y = µ + G + S + A + G × S + G × A + S × A + G × S × A + ε, where G (genotype, mutant, or control), S (sex, male [M] or female [F]), and A (age, week 0 or 3) are main effects, Y is the enzyme activity, µ is the overall mean, and ε is the residual error.

HS isolation and LC–MS quantification

Total HS levels were analyzed for 7 biological replicates (3–5-day-old flies) for each genotype and sex, with 5 whole flies in each biological replicate. Lyophilized fly samples were preweighed and then homogenized with a pestle, and sonicated in a 50-mM sodium acetate buffer, pH 6.0, containing 200 mM NaCl. Subsequently, samples were diluted 1:10 in buffer containing 0.1% Triton X-100 and Pronase (0.4 mg/mL, Sigma cat# P5147) and digested overnight at 37°C with mild agitation. The product was centrifuged, and the cleared supernatant was passed through a DEAE-Sephacel (Cytiva) column equilibrated in 50 mM sodium acetate buffer, pH 6.0, containing 200 mM NaCl and then passed through a PD-10 desalting column (Cytiva). For HS disaccharide analysis, lyophilized glycosaminoglycans (GAGs) were digested with heparin lyases I, II, and III (2 mU each, IBEX) for 16 h at 37°C in a buffer containing 40 mM ammonium acetate and 3.3 mM calcium acetate, pH 7. Enzymatically depolymerized HS samples were differentially mass labeled by reductive amination with ¹²C-aniline, as previously reported (Lawrence et al. 2008). HS disaccharides were quantified via hydrophilic interaction liquid chromatography–mass spectrometry (HILIC LC–MS) by comparison with a set of commercial HS disaccharide standards (Iduron) that were tagged with ¹³C-isotopically labeled aniline and spiked into each sample. Tagged HS disaccharides were analyzed on a Waters SYNAPT XS mass spectrometer equipped with an ACQUITY UHPLC H-class system (BEH Glycan Column, 2.1 × 100 mm). Mobile phase A was 50 mM ammonium formate buffer, pH 4.4. Mobile phase B was 100% acetonitrile. The elution was as follows: 0–5.0 min with isocratic 10% A, linear gradient 10–33% A for 5.0–49.0 min, linear gradient 33–45% A from 49.0 to 51.5 min, isocratic 45% A from 52.0 to 60 min, and then the column was washed with 90% B for 10 min. The column temperature was room temperature, and the flow rate used was 0.5 mL/min (injection volume of 2 µL). The accumulative extracted ion current was computed, and further data analysis was carried out using the MassLynx software suite. Total HS levels were normalized to dry sample weight, and pairwise analyses between mutant and control was performed using the fixed effects ANOVA model, Y = µ + G + S + G × S + ε, where G (genotype, mutant, or control) and S (sex, M or F) are main effects, Y is the HS levels, µ is the overall mean, and ε is the residual error.

Lysotracker staining and imaging

LysoTracker Green DND-26 (1 mM; ThermoFisher Scientific, Inc.) was diluted 1:2,000 in PBS. Adult fly brains (weeks 0 and 3) and ventral nerve cords (VNCs, week 0) were dissected in PBS and immediately transferred to a glass slide containing 12 µL Lysotracker solution. Imaging for all samples was performed within 15 min of dissection using a Keyence microscope (BZ-X800E, Keyence, Raleigh, NC, USA) at a 10× objective lens with a constant exposure time. Twenty 1 µm slices were captured and stacked into a single z-projection image. Acquired images were blinded and quantified using ImageJ (Schindelin et al. 2012). Total brain area and total VNC area were captured using the polygon tool, and the area of stained puncta was calculated using the “analyze particle” function, which was then used to calculate the percentage of stained area. Pairwise analyses for Lysotracker staining of brain samples between mutant and control were performed using a fixed effects ANOVA model, Y = µ + G + S + A + G × S + G × A + S × A + G × S × A + ε, where G (genotype, mutant, or control), S (sex, M or F) and A (age, week 0 or 3) are main effects, Y is the percent area of puncta measured, µ is the overall mean, and ε is the residual error. Pairwise analyses for VNC staining were performed using the same model without the age term.

Sleep and activity

We used the Drosophila Activity Monitoring System (DAM system, TriKinetics, Waltham, MA, USA) to measure sleep and activity phenotypes. Young flies (3–5 days old) and 20–22-day-old flies were placed in DAM tubes containing 2% agar with 5% sucrose and sealed with rubber caps (TriKinetics) and a small piece of yarn. Measurements were recorded for a 7-day period on a 12-h light–dark cycle. These measurements were separated into a 5-day period after discarding the first day and last day of recordings. Four full-monitor replicates (n = 128) were performed for each genotype/sex/age. Flies that did not survive the full recording period were discarded from analyses. The raw data were first processed using an in-house pipeline to remove any aberrations in the recording time (recordings taken over or under one minute). Filtered data were then processed using ShinyR-DAM (Cichewicz and Hirsh 2018), and the resulting output files were used to generate sleep and activity phenotypes. Individual fly day–night sleep data, individual sleep activity bout data, and daily locomotor activity data files were used to generate datasets for sleep, activity, and locomotion profiles using an in-house processing pipeline (https://github.com/ypan23-1876660/DAM). Pairwise analyses between mutant and control were performed using a mixed effects ANOVA model, Y = µ + G + S + A + G × S + G × A + S × A + G × S × A + Rep(G × S × A) + ε, where G (genotype, mutant, or control), S (sex, M or F), and A (age, week 0 or 3) are main effects, G × S, G × A, S × A, and G × S × A are interaction terms, Rep is the random effect of replicate monitor, Y is the sleep or activity phenotype, µ is the overall mean, and ε is the residual error.

RNA sequencing

Adult fly brains were dissected between 9 AM and 12 PM in cold PBS. Dissections were randomized for all genotypes and sexes and performed separately for each age group (weeks 0 and 3), with 24 samples for each age group. Six samples were collected each day. For each sample, 20 brains were collected in a 1.7-mL centrifuge tube and flash-frozen on dry ice. Brains were lysed using metal beads, and bulk RNA was extracted using a modified Zymo Research Direct-zol RNA Microprep kit protocol. RNA libraries were prepared with a Tecan Genomics Universal Plus system. We sequenced 3 biological replicates per genotype for each sex and each age for a total of 48 samples using a single S1 flow cell on the NovaSeq 6000 platform (Illumina, Inc.).

We used the fastP pipeline (Chen et al. 2018) to filter out short and low-quality reads, and quantified rRNA using the bbduk command. The reads were then aligned to the Drosophila reference genome (FlyBase Release 6.13). Read counts were generated using the feature counts pipeline from the Subread package (Liao et al. 2014). Genes with a median expression of log₁₀ ≤ 2 or with an expression of 0 in more than 75% of the samples were excluded from the analyses. We normalized the data using the gene length corrected trimmed mean of M values (Ge-TMM) method (Smid et al. 2018) before performing differential expression analysis.

Differential gene expression analyses and interaction network construction

Differential gene expression analyses were performed using a fixed effect full-factorial ANOVA model, Y = µ + G + S + A + G × S + G × A + S × A + G × S × A + ε, where G (genotype, mutant, or control), S (sex, M or F), and A (age, week 0 or 3) are main effects and G × S, G × A, S × A, and G × S × A are the interaction terms, Y is gene expression, µ is overall mean, and ε is residual error. We used the same model to perform pairwise analyses between each mutant and control for the genes that were significantly differentially expressed (false discovery rate [FDR] <0.05) in the initial analysis.

Known genetic and physical interactions of all genes in the fly genome were acquired from the FlyBase database (Öztürk-Çolak et al. 2024). We used Cytoscape (V3.9.1) to construct a global interaction network with all known genetic and physical interactions in the fly genome (Shannon et al. 2003). We applied a “column filter” to this global interaction network to select nodes representing our differentially expressed genes. Next, we applied a “topology filter” to select nodes that interact with at least 2 differentially expressed genes.

Statistical analyses

All statistical analyses for biochemical, neuroanatomical, and behavioral data were performed in SAS Viya V.03.05 (SAS, Cary, NC, USA) using the “PROC MIXED” command to test the type III factorial ANOVA. Statistical analyses for transcriptomic data were performed using the “PROC GLM” command to test the type III ANOVA, and the Benjamini–Hochberg procedure (Benjamini and Hochberg 1995) was applied in R to correct for multiple hypothesis testing. For biochemical, neuroanatomical, and behavioral analyses, P < 0.05 was used to infer significance, and for transcriptomic analyses, Benjamini–Hochberg corrected P < 0.05 was used to infer significance.

Results

Effects of Naglu mutations on enzyme activity and HS levels

We measured enzyme activity for all genotypes at young (week 0) and old (week 3) ages, separately for males and females. For both sexes, we observed significantly reduced enzyme activity across all mutants relative to the control (P < 0.0001 for all mutants; Fig. 1a and b; Supplementary Table 1). Residual enzyme activity was further reduced with age across all mutations.

Fig. 1.

Effects of mutations on enzyme activity and HS levels. a) Bar graphs showing Naglu enzyme activity ± SE in females. b) Bar graphs showing Naglu enzyme activity ± SE in males. c) Bar graphs showing total HS levels ± SE based on LC–MS disaccharide analysis data in females and males for 3–5-day-old flies. For a) and b), n = 3; for c), n = 7. Asterisks in a) and b) indicate comparisons of each genotype with the control (sexes separately, ages pooled), and asterisks in c) indicate comparisons of each genotype with the control (sexes separately). **P < 0.01, ***P < 0.001, ****P < 0.0001.

Next, given the significant reduction of enzyme activity levels in young flies, we tested the effects of mutations on HS levels at week 0. Evaluation of total HS residues on whole flies using LC–MS showed significant increases in total HS in all mutants, Naglu^KO, Naglu^Y160C, and Naglu^W422X (P < 0.0001 for all, sexes pooled), for both males (P = 0.0001, P < 0.0001, and P = 0.0001, respectively) and females (P < 0.0001, P < 0.0001, and P = 0.0001, respectively) (Fig. 1c; Supplementary Table 2). Thus, we observe a similar reduction in enzyme activity levels across all mutants with a concomitant increase in HS accumulation.

Effects of Naglu mutations on acidic compartments in the brain and VNC

Accumulation of HS results in increased lysosomal volumes in MPS IIIB patients. Lysotracker is a basic dye that stains acidic compartments, including lysosomes. We stained and imaged whole mount fly brains of mutant and control flies with Lysotracker and quantified Lysotracker fluorescence to measure the disruption of acidic compartments. Positively stained areas were quantified as a percent of the total brain area. Mutants exhibited enlarged fluorescent puncta that were not observed in the controls (Fig. 2a–d). Fluorescent puncta were especially concentrated in the optic lobe (Fig. 2b–d, right panels) and around the Kenyon cells of the mushroom body. Mutants of both sexes exhibited significantly higher levels of staining compared with the controls (P < 0.0001 for both pooled and separate analyses; Fig. 2e and f; Supplementary Table 3). This effect was more pronounced with age. However, Naglu^W422X males showed a higher level of Lysotracker staining at week 0 than at week 3 (age effect, P = 0.0357).

Fig. 2.

Lysotracker staining of fly brains. a–d) Representative whole mount brains at 100× magnification on the left and 400× magnification of the boxed area of the optic lobe on the right. a) Naglu⁺, b) Naglu^KO, c) Naglu^Y160C, and d) Naglu^W422X. e and f) Bar graphs showing average percent area staining of whole mount brains ± SE 100× magnification. e) Females and f) males. n = 11–13 for each sex, genotype, and age. Asterisks indicate comparisons of each genotype with the control (sexes separately, ages pooled). ****P < 0.0001.

To further investigate the effects of Naglu mutations on the CNS, we stained and imaged VNCs of mutant and control flies with Lysotracker. The VNC is responsible for processing motor signals from the brain to generate coordinated locomotor outputs (Venkatasubramanian and Mann 2019). Given the significant disruption of acidic compartments observed in the brains of young flies, we only stained VNCs at week 0. At this time point, mutants presented substantially enlarged fluorescent puncta compared with the control (Fig. 3a–d). Mutants of both sexes had significantly increased levels of Lysotracker staining relative to the control (Fig. 3e; Supplementary Table 4).

Fig. 3.

Lysotracker staining of the fly VNC. a–d) Representative VNCs at 100× magnification on the left and a boxed area of 600× magnification of neuropil on the right. a) Naglu⁺, b) Naglu^KO, c) Naglu^Y160C, and d) Naglu^W422X. e) Bar graphs showing the average percent area staining of VNC ± SE for females and males at 100× magnification. n = 6–7 for each sex and genotype. Asterisks indicate comparisons of each genotype with the control (sexes separately). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Effects of Naglu mutations on sleep and activity

Hyperactivity and sleep disturbances are characteristic of the progression of MPS IIIB in human patients as one of the earliest phenotypes. We quantified sleep and activity in mutant and control flies at young (week 0) and older (week 3) ages (Fig. 4). We observed an overall decrease in total night sleep (time spent sleeping in a 12-h dark cycle) across all mutations relative to the controls when analyzed separately for sexes and ages (Fig. 4a; Supplementary Table 5). When pooled across sexes and ages, we observed higher significance for decreased total night sleep in all 3 mutants, Naglu^KO, Naglu^Y160C, and Naglu^W422X (P < 0.0001, P = 0.0014, and P = 0.0143, respectively), relative to controls (Supplementary Table 5). Next, we analyzed the nighttime sleep bout counts (the number of times sleep is initiated in a 12-h dark cycle) as a measure of sleep fragmentation. While there was a trend of sleep fragmentation across all genotypes (excluding Naglu^W422X females), analyses for fully reduced models (reduced by both age and sex) yielded significant results only for Naglu^KO males at week 3 (P = 0.0011) and Naglu^Y160C females at week 0 (P = 0.0068; Fig. 4b; Supplementary Table 5). Data pooled across sexes and ages showed that Naglu^KO and Naglu^Y160C had significantly higher sleep fragmentation relative to controls (P < 0.0001 and P = 0.0084, respectively); however, Naglu^W422X did not show significantly different sleep fragmentation (P = 0.4426; Supplementary Table 5).

Fig. 4.

Box plots showing sleep and activity profiles of Naglu mutant flies and the control. a) Total night sleep, b) nighttime sleep bout count, c) total locomotor activity, and d) total activity bout length. n = 112–128 flies per genotype per sex per age. Asterisks indicate pairwise comparisons of each genotype with the control. Each mutant for each sex and age is compared with the control for the same sex and age. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. F, female; M, male.

Reduced sleep was accompanied by hyperactivity for all mutations. We observed sex- and age-specific effects for locomotor activity (the number of times flies traversed the infrared beam in a 24-h period), where Naglu^KO, Naglu^Y160C, and Naglu^W422X females exhibited severe hyperactivity at week 3 (P = 0.0005, P = 0.0033, and P = 0.0019, respectively; Fig. 4c; Supplementary Table 5). When pooled across both sexes and both ages, Naglu^KO, Naglu^Y160C, and Naglu^W422X flies had significantly higher total locomotor activity relative to control (P < 0.0001, P = 0.0003, and P < 0.001, respectively; Supplementary Table 5).

Next, we analyzed total activity bout length (the average length of each activity bout in a 24-h period). We observed significantly higher total activity bout length relative to controls for Naglu^KO females at week 3 (P < 0.0001) and Naglu^KO males at week 0 (P = 0.0007). Similarly, we found significantly higher total activity bout length relative to controls for Naglu^W422X females at week 3 (P = 0.031) and Naglu^W422X males at week 0 (P = 0.0004). Interestingly, Naglu^W422X males had significantly reduced total locomotor activity at week 3 (P = 0.0018; Fig. 4d; Supplementary Table 5). When pooled across both sexes and both ages, Naglu^KO and Naglu^W422X had a significantly higher activity bout length (P < 0.0001 and P = 0.0008, respectively), but not Naglu^Y160C (P = 0.4850; Supplementary Table 5). These results show that Naglu mutations affect sleep and activity consistent with MPS IIIB phenotypes.

Effects of Naglu mutations on whole-genome gene expression

To identify genes and pathways perturbed by Naglu mutations, we performed whole-genome gene expression analyses on brain samples of all mutant genotypes and the control across both sexes and both ages. Filtering for low-quality reads and transcripts with low expression (median log₁₀[counts] ≤ 2) resulted in 12,028 transcripts across all samples. We tested the fixed effects of genotype, sex, age, and their interaction on the expression of each gene. At a FDR < 0.05, we identified 192 genes with a significant effect of genotype, 1,849 genes with a significant effect of sex, 5,910 genes with a significant effect of age, and 41 genes with a significant sex × age interaction term (Supplementary Table 6). At FDR < 0.05 or FDR < 0.1, there were no genes with significant genotype × sex, genotype × age, and genotype × sex × age interaction terms.

We performed pairwise analyses between each mutant and control for the 192 genes that were differentially expressed across genotypes. Among these genes, 111 were differentially expressed for Naglu^KO (57 downregulated and 54 upregulated), 109 genes were differentially expressed for Naglu^Y160C (46 downregulated and 63 upregulated), and 109 genes were differentially expressed for Naglu^W422X (48 upregulated and 61 downregulated), and 33 genes were common among all 3 mutants (14 upregulated and 19 downregulated). Thirteen genes that were differentially expressed in more than one mutant had effects in opposite directions. Approximately 35 genes were specific to each genotype (Fig. 5). Interestingly, 25 snoRNAs (24 upregulated and 1 downregulated) and 3 lncRNAs (all downregulated) were differentially expressed only in Naglu^Y160C (Supplementary Table 7d). Additionally, the expression of Naglu was significantly downregulated in all 3 mutants; however, the level of downregulation varied across genotypes, with Naglu^KO showing virtually no expression and Naglu^Y160C showing the highest expression among mutants (Supplementary Fig. 2).

Fig. 5.

Differentially expressed genes. Venn diagram showing the number of differentially expressed genes for each Naglu mutant relative to the control.

Brain transcriptome profiles

To assess the biological relevance of differentially expressed genes across all mutations, we performed gene ontology (GO) analysis of the 192 differentially expressed genes and observed no significant enrichment in GO categories or Kyoto Encyclopedia of Gene and Genomes pathways. However, differentially expressed genes in one or more mutants were associated with different aspects of nervous system development, including gene products associated with neural regulation, signaling, proteolysis, vesicle trafficking, mitochondrial function, and chromatin modification. 5-HT1A was downregulated in all mutants and has been implicated in sleep regulation (Yuan et al. 2006). bsk, downregulated in all mutants, has been implicated in mushroom body development (Bornstein et al. 2015) and axon guidance (Srahna et al. 2006; Soares et al. 2014). Notably, transcripts associated with regulation of the Notch signaling pathway, including numb (Guo et al. 1996), insv (Duan et al. 2011), Elba2 (Ueberschär et al. 2019), and bib (Doherty et al. 1997) show altered regulation in mutant backgrounds along with transcripts that regulate actin and cytoskeleton organization, including Arpc1 (Stevenson et al. 2002), chic (Baum and Perrimon 2001), and CAP (Stevenson and Theurkauf 2000). Kermit, which interacts with a guanyl cyclase to regulate axon guidance (Chak and Kolodkin 2014), Rchy1 that regulates gliogenesis (Foo et al. 2017), C1GalTA, which encodes a galactosyl transferase A (Lin et al. 2008), and babos, which is a cell adhesion protein and contributes to synaptic plasticity (Bai and Suzuki 2020), also show altered expression when Naglu is disrupted. Eogt, which encodes EGF-domain-O-N-acetylglucosamine transferase, is downregulated in all 3 mutants. This enzyme is located in the lumen of the endoplasmic reticulum where it contributes to protein glycosylation by catalyzing the transfer of GlcNAc from UDP-GlcNAc to serine and threonine hydroxyl groups of proteins (Sakaidani et al. 2011).

To assess functional contexts of the differentially coregulated genes, we used known genetic and physical interactions in the FlyBase database (Öztürk-Çolak et al. 2024) to construct interaction networks with 1st-degree neighbor interactions between at least 2 of the 192 differentially expressed genes (Fig. 6). These networks revealed 2 distinct subclusters on which we performed GO enrichment analyses. The 1st subcluster, anchored by Snap29 and Syb, was enriched for GO terms associated with vesicle trafficking and vesicle-mediated transport of neurotransmitters (Fig. 6; Supplementary Fig. 3a). The 2nd subcluster was enriched for GO terms associated with development, including development and function of the nervous system (Fig. 6; Supplementary Fig. 3b).

Fig. 6.

Differentially expressed genes and their interacting partners. The interaction network was built from 192 differentially expressed genes and their known genetic and physical associations.

Discussion

MPS IIIB is an autosomal recessive disorder caused by mutations in NAGLU and has severe effects on the CNS. Here, we developed a fly model of the disorder using mutations in Naglu that are orthologous to patient mutations. Naglu mutants of both sexes have significantly reduced enzyme activity accompanied by increased levels of HS. Lysotracker staining shows the accumulation of fluorescent puncta in the brain, likely reflecting swelling of lysosomes due to the accumulation of HS breakdown products, which may ultimately lead to disruption of the autophagic pathway and cellular dysfunction (Monaco and Fraldi 2021). The same pattern of Lysotracker staining was observed in the VNC. Additionally, mutant flies exhibit behavioral hallmarks of MPS IIIB such as sleep defects and hyperactivity. Although we did not observe loss of motor functions in our flies, disruption of acidic compartments in VNC could implicate downstream effects in motor coordination, including the hyperactivity observed in mutant flies (Fig. 4c and d). Sleep alterations show a significant decrease in overall night sleep with an increase in sleep fragmentation (Fig. 4a and b). This is accompanied by higher activity during the day, further indicating the lack of sleep rebounds. We also observe sex- and age-dependent effects. Such context-dependent effects are common for virtually all phenotypes modeled in Drosophila, e.g. longevity (Huang et al. 2020), sleep (Harbison et al. 2013), and starvation resistance (Harbison et al. 2004). We did not observe late-onset loss of locomotion or effects on longevity in our Drosophila model. It is likely that the relatively short lifespan of Drosophila prevents the manifestation of the long-term effects of Naglu mutations seen in human patients. We did not assess the effects of Naglu mutations in flies older than 3 weeks, since observations of decline in locomotor activity at older ages would be confounded by naturally occurring decline in locomotion because of senescence.

Although all mutants were maintained in the same genetic background, the severity of the assessed traits varied across mutations. The deletion mutant Naglu^KO and the truncation mutant Naglu^W422X had large effects, whereas the missense mutant Naglu^Y160C had less severe sleep and activity defects. The disruption of whole brain acidic compartments was less severe for Naglu^Y160C as well. Indeed, compared with Naglu transcript levels in the control, examination of transcript expression shows the virtual absence of the Naglu^KO transcript, significant reduction of the Naglu^W422X transcript, and a small reduction in the transcript level of Naglu^Y160C (Supplementary Fig. 2). However, measurement of enzyme activity and HS accumulation indicated a similar reduction in Naglu activity (Fig. 1a and b) and increase in HS accumulation (Fig. 1c) across all mutations. It is possible that the translation of the Naglu^Y160C transcript into an active enzyme is compromised resulting in reduced enzyme activity similar to that of the Naglu^KO and Naglu^W422X gene products despite a smaller reduction in gene expression. Residual enzyme activity observed in the Naglu^KO mutant (Fig. 1a and b) may reflect background noise due to nonspecific fluorescence in the assay unrelated to Naglu activity.

Transcriptional profiling of fly brains identified differentially expressed genes in Naglu mutants in a variety of biological processes that affect the development of the nervous system. Numb, downregulated in Naglu^KO and Naglu^Y160C, and insv, downregulated in all mutants, are negative regulators of the Notch signaling pathway. numb inhibits Notch during nervous system development to regulate cell fate determination (Guo et al. 1996). insv also negatively regulates Notch signaling by corepressing its activator Su(H) (Duan et al. 2011). We also observed the downregulation of Eogt, which mediates O-glycosylation in the endoplasmic reticulum by transferring GlcNAc to serine and threonine residues of proteins. An increase in extracellular HS has been observed in MPS III studies (Bruyère et al. 2015; De Pasquale et al. 2021). In contrast to the behavioral assays, the transcriptomic analyses did not show the effects of sex or age on differential gene expression across the mutants. Consistent with the transcriptome data, Naglu^Y160C has less severe behavioral phenotypes (Fig. 4), fewer differentially expressed mRNAs, and more differentially expressed snoRNAs compared with the other 2 mutants (Fig. 5).

We constructed an interaction network from differentially expressed genes and their known genetic and physical interacting partners. This network contains 2 subnetworks (Fig. 6). Subnetwork 1 is enriched for genes associated with vesicle trafficking and synaptic vesicle release, including members of the SNARE family which function in membrane fusion (Supplementary Fig. 3a). Snap29, upregulated in Naglu^W422X, was also upregulated in a proteomics study that examined the brains of NAGLU^−/− mice (Petrova et al. 2023). It is involved in autophagic regulation, autophagosome-lysosome fusion, synaptic transmission, and cell division (Mastrodonato et al. 2018). Another member of the SNARE family, Synaptobrevin (Syb), which is downregulated in Naglu^KO and Naglu^Y160C, is involved in neurotransmitter release. Loss of Syb in Drosophila leads to neurodegeneration (Haberman et al. 2012). Our observations and previous studies on MPS IIIA (Settembre et al. 2008) indicate that lysosomal dysfunction that arises from disruption of HS breakdown has widespread effects on the steady state of intracellular vesicle transport including synaptic vesicles.

Subnetwork 2 is enriched for genes with GO terms associated with nervous system development (Supplementary Fig. 3b). Although not included in this subnetwork, we found that the serotonin receptor, 5-HT1A, was downregulated in all mutants. Loss-of-function mutations in 5-HT1A reduce total sleep and sleep bout lengths (Yuan et al. 2006). Decreases in global serotonin levels have also been reported in MPS IIIA studies (Fu, Meadows, Pineda et al. 2017; Fu, Meadows, Ware et al. 2017). Additionally, the use of the selective serotonin reuptake inhibitor fluoxetine improves autophagic-lysosomal functions in MPS IIIA cellular and mouse models (Capuozzo et al. 2022). We hypothesize that dysregulation of cell fate determination during the early stages of development leads to changes in neural circuitry which affect behavioral phenotypes.

MPS IIIA and MPS IIIB are highly penetrant and manifest aggressively in most cases (Ouesleti et al. 2011). However, there have been reports of MPS IIIA patients with milder forms of the disorder, and some of these have been linked to specific mutations (Meyer et al. 2008). However, some studies have also suggested that people in certain geographic regions are subjected to less severe forms of MPS IIIB (Valstar et al. 2010). This indicates the potential for exploring the effects of genetic background on the severity of the disorder, which can be pursued in the Drosophila model. About 75% of human disease–associated genes have Drosophila orthologs, which enables translational inferences from Drosophila to human populations (Pandey and Nichols 2011). Thus, establishing a Drosophila model for MPS IIIB can provide an in vivo system for the development of new therapeutics targeting coregulated genes associated with Naglu mutations or genetic modifiers that ameliorate Naglu mutant phenotypes.

Data availability

Biochemical data, behavioral data, Lysotracker data, RNA-seq combined counts, and coding scripts are available on GitHub at https://github.com/bsimkha/Dmel_MPSIIIB_model. RNA sequencing data have been deposited in the Gene Expression Omnibus (GEO) database under accession number GSE269032. Mutant fly stocks used in the experiments will be made available upon request. The original stocks created by WellGenetics are available from the Bloomington Drosophila Stock Center (Bloomington, IN, USA).

Supplemental material available at GENETICS online.

Acknowledgments

We thank Dr Richard Steet for helpful discussions. We thank the Genomics and Bioinformatics Cores of the Clemson University Center for Human Genetics and Dr Parastoo Azadi and the UGA Complex Carbohydrate Research Center Analytical Services team for the use of resources and assistance.

Funding

This work was supported by a grant from the Cure Sanfilippo Foundation and, in part, by the Genomics and Statistics and Bioinformatics Research Core of National Institutes of Health grant 1P20GM139769-01 to TFCM and RRHA. The Cure Sanfilippo Foundation also funded the creation of the original MPS IIIB fly model by WellGenetics, Inc. (Taipei, Taiwan). HF-S is supported by National Institutes of Health grant R01NS128907. RJW is supported by National Institutes of Health grant R35GM150736 and a grant from the Cure Sanfilippo Foundation. The Analytical Services and Training Core at the University of Georgia that provided instrumentation to perform the glycosaminoglycan analyses is supported by National Institutes of Health grant R24GM137782.

Author contributions

BS performed the experiments and analyzed the results, NON-Y designed fly crosses and maintained fly stocks, PSF assisted in dissecting brains and Lysotracker staining, RAL performed RNA extractions and sequencing, VS provided bioinformatics support, KW and HF-S performed the Naglu enzyme assays, AB and RJW performed LC–MS analyses of HS, RRHA and TFCM conceptualized and supervised the project and provided resources. BS, RRHA, and TFCM wrote the manuscript.

Author notes

Conflicts of interest: The author(s) declare no conflicts of interest.